Skip to main content
Raigal catches the sloppy patterns AI coding tools leave behind: redundant comments, swallowed errors, unsafe type casts, and insecure calls. It packages linting, formatting, security checks, and pattern detection into one CLI command with a calibrated score from 0 to 100. Connected to Raigal Cloud and the official Raigal GitHub App, it evaluates pull requests, protects main branches, tracks issues on a 4-stage board, and coordinates automated fixes with local coding agents.
Your source code is never uploaded to Raigal servers. During pull request checks, an in-memory secret-masked patch diff (under 20 KB) is sent to display changes in the dashboard.

​
What Raigal catches

AI assistants like Claude Code, Cursor, Copilot, and OpenCode speed up development, but frequently leave behind structural defects:
  • Narrative comments that repeat function names or obvious code.
  • Empty catch blocks that hide runtime failures.
  • Unsafe type assertions (as any, as unknown as T) that bypass the compiler.
  • Leftover debugging statements, console logs, and unreachable branches.
  • Unsanitized SQL, dynamic eval(), and unsandboxed agent tools.
Raigal runs through local agent hooks, in CI on every pull request, and during terminal audits.

​
Components

​
Getting started

1

1. Sign in to Raigal Cloud

Create a workspace at app.raigal.dev with a 14-day evaluation trial and invite your team. Organization onboarding guide
2

2. Install the Raigal GitHub App

Connect your repositories with the Raigal GitHub App to run checks on pull requests. Pull request checks
3

3. Add the CI workflow

Add a GitHub Actions workflow to check pull requests and send scores to your dashboard. CI/CD pipelines
4

4. Set up local tools and hooks

Install the CLI (raigal scan, raigal fix) and set up interception hooks for Claude Code or Cursor. Quickstart guide